A token buyback uses assets to purchase a project's own tokens. A burn permanently removes token units under the relevant network or contract rules. Bought tokens can remain spendable in a treasury, and burns can use tokens already held without a purchase. Verify the funding source, destination, burn mechanism and net supply change separately.
A hypothetical headline such as “$50 million returned to holders” leaves several questions open. Was that an approved budget, completed purchases or the market value assigned to destroyed tokens? Did the acquired tokens enter a reserve that can distribute them again? Did new issuance exceed the burn during the same period?
Follow two records together: an asset ledger showing what paid for purchases, and a token ledger showing ownership and supply. Combining those records prevents a purchase followed by a burn from becoming two separate claims of value delivered.
Four operations behind similar headlines
A buyback creates a purchase transaction. The buyer gives up a settlement asset and receives tokens. Once acquired, the tokens might be held, distributed, locked or burned. The purchase alone does not determine their eventual treatment.
Aave provides a documented example of buying and retaining tokens. Its official documentation describes protocol-revenue-funded AAVE purchases going to the DAO's Ecosystem Reserve for governance-approved programs, including rewards, grants and service provider payments. It explicitly describes this destination as an alternative to burning. This is evidence about the documented mechanism; it does not establish purchases on any particular day. Aave's AAVE token documentation.
A burn can also happen without buying tokens. Ethereum's EIP-1559 specifies that the transaction base fee is burned. That is a network fee mechanism, with users supplying the ETH through transaction fees. Its burn record does not describe a treasury purchasing ETH in the market. EIP-1559's base-fee specification.
| Operation | What moves or changes? | What remains to verify? |
|---|---|---|
| Buy and hold | Settlement assets leave; purchased tokens enter a controlled account | Spending authority, later distributions and any float classification |
| Buy and burn | A purchase occurs, then acquired inventory is retired | Both transaction stages, quantities and the burn implementation |
| Burn existing inventory | Previously held token units are retired | Inventory origin and whether it was counted as circulating |
| Fee burn | A fee rule removes units paid through network or application activity | The fee component, covered activity and concurrent issuance |
These mechanisms can coexist in one project. Keep their quantities separate before calculating a combined total. For the earlier step of identifying the cash available to the protocol, the fees versus revenue guide traces user payments through suppliers, treasury allocations and costs.
Reconcile purchases, inventory and cash
Consider a hypothetical token over one closed month. This fictional accounting exercise assumes a single chain, ordinary non-rebasing balances and a burn function that reduces total supply. There are no bridges, transfer taxes or other changes beyond those listed.
The token starts with 1,000,000 units: 200,000 in the treasury and 800,000 held outside it. For this example only, all outside balances count as circulating and treasury balances do not. This explicit convention lets the exercise track circulation without claiming that any data provider uses the same rule.
The treasury starts with $20,000 of settlement assets and receives $30,000 of current-period protocol receipts. Assume those assets remain worth exactly $1 per unit. A $50,000 spending authorization covers purchases and their execution costs.
Three completed purchase tranches acquire 10,000 tokens at $2, 5,000 at $2.40 and 5,000 at $2.60. The costs are $20,000, $12,000 and $13,000. Purchases therefore total 20,000 tokens for $45,000, giving a volume-weighted acquisition price of $2.25 per token before execution costs.
Separate network and execution expenses total $1,000, paid from the same settlement-asset balance. Actual spending is $46,000 and closing settlement assets are $4,000: opening $20,000 plus receipts of $30,000 minus spending of $46,000. The $50,000 authorization is neither the purchase amount nor proof of expenditure.
Now add the token events in order:
| Hypothetical event | Treasury tokens after event | Outside tokens after event | Total supply after event |
|---|---|---|---|
| Opening balances | 200,000 | 800,000 | 1,000,000 |
| Buy 20,000 from outside holders | 220,000 | 780,000 | 1,000,000 |
| Burn 55,000 treasury tokens | 165,000 | 780,000 | 945,000 |
| Pay a 10,000-token grant from treasury | 155,000 | 790,000 | 945,000 |
| Mint 25,000 rewards to outside holders | 155,000 | 815,000 | 970,000 |
For inventory attribution, stipulate that 15,000 burned units are assigned to this month's acquisitions and 40,000 to opening treasury inventory. The remaining 5,000 acquired units stay in treasury. Tokens are fungible: this is a declared accounting attribution, rather than proof that individual token units retain identities.
The acquisition record reconciles as 20,000 bought equals 15,000 assigned to burns plus 5,000 retained. The broader treasury record reconciles as 200,000 opening plus 20,000 acquired minus 55,000 burned minus 10,000 distributed equals 155,000 closing.
Gross burns are 55,000 tokens, or 5.5% of opening total supply. Net total-supply reduction is only 30,000, or 3%, because 25,000 new tokens are issued. Adding the 20,000 purchased units to the 55,000 burned units would produce an unsupported 75,000-unit “supply removal” claim: the purchase changed ownership, while the burn changed total supply.
Under our circulation convention, outside holdings rise by 15,000, from 800,000 to 815,000, or 1.875%. Thus total supply shrinks while measured circulating supply grows. The free float versus circulating supply comparison explains why real-world exclusions require their own ownership evidence and methodology.
The funding record also limits the headline. Current receipts contribute $30,000 toward $46,000 spent, so settlement assets decline by $16,000. The program used accumulated resources as well as current receipts. Repeating the same spending level would require future receipts, additional reserves or another funding source; this example supplies no evidence about those future amounts.
Establish what the burn actually changed
In OpenZeppelin Contracts 5.x, the standard ERC-20 implementation distinguishes ordinary transfers from its internal burn operation. Its burn operation lowers total supply and emits a transfer event to the zero address. Ordinary transfers to the zero address are rejected. This implementation illustrates why a burn function and a destination-address transfer require different checks. OpenZeppelin's ERC-20 implementation reference.
A transfer to an address described as “dead” may leave the contract's reported total supply unchanged. Assess spendability under the actual rules and preserve the address classification evidence. An explorer label alone does not explain the supply calculation. CoinGecko, for example, describes total supply as on-chain supply minus burned tokens and defines its burner-address criterion in terms of inaccessible funds. Record that provider adjustment separately from the contract's own value. CoinGecko's supply methodology
Read the relevant implementation, transaction receipt and before-and-after state at recorded blocks. For a conventional supply-reducing burn, reconcile the holder's balance reduction, burn quantity and supply reduction, allowing for other minting or burning in the comparison interval. Unexplained differences require investigation before publication.
Retirement of existing units also leaves a separate question: can authorized actors create more units later? The ability to mint or change the implementation affects future supply even when the recorded burn is valid. Distinguish completed destruction from a claim that supply can never expand again.
Check the economic boundary
A local burn may be part of moving a token between networks. Ethereum's bridge documentation describes burn-and-mint designs in which source-chain units are burned and destination-chain units are minted. Counting only the first leg would misstate the combined change across those representations. Ethereum's bridge mechanism documentation.
Redemption can give a burn another purpose. ERC-4626, for example, specifies vault withdrawal and redemption operations that burn shares and send underlying tokens to a receiver. This cancels a claim while releasing assets. ERC-4626's tokenized-vault standard.
Staking receipt implementations require their own rules. The liquid staking token discount guide follows the separate relationship between receipt units, underlying units and market price. A burn transaction needs its surrounding purpose before it becomes evidence of a value-distribution mechanism.
Compare money spent with tokens retired
Keep three numbers in their original units: settlement assets paid, tokens acquired and tokens burned. Dollar burn values are valuations, while purchase spending is a transaction amount.
Suppose the hypothetical month's 55,000 burned tokens are valued at a stipulated month-end price of $3. The resulting $165,000 burn value does not mean the treasury spent $165,000. Its purchases cost $45,000, and most of the burned inventory was already held. Revaluing the same burn at $4 would produce $220,000 without another purchase or another token being destroyed.
If purchased tokens are later burned, recording acquisition spending and retired quantity is useful. Adding their dollar valuations as two independent payments to holders double counts stages of the same mechanism. Holders who keep their tokens receive no automatic cash distribution from this accounting sequence.
A 3% supply reduction is also not a forecast of a 3% price increase. It describes a unit count. The ledger supplies no assumption that market demand, available liquidity or other holders' selling remains fixed.
The funding asset matters too. A program can spend fees, accumulated reserves, borrowed assets or proceeds from selling another treasury asset. Each source has different obligations and remaining resources. A “revenue-funded” label needs a reconciliation showing the relevant receipts and deductions; gross user fees do not establish spendable revenue.
Purchase evidence needs more than incoming token transfers. Match settlement-asset debits to acquired quantities, fees, venue records or swap execution, then identify the receiving controller. A transfer between two treasury addresses is an internal movement. The token velocity guide shows why repeated transfers and economic transactions require different interpretations.
Build an evidence packet for one announcement
Use a closed period and retain the fields below. A proposal establishes authorization or intent; completed execution requires its own records.
| Field | Evidence to retain | Question it resolves |
|---|---|---|
| Identity and window | Chain, contracts, controlled accounts, block range and UTC dates | Which assets and events are covered? |
| Funding | Opening settlement balances, receipts, borrowing and asset sales | What resources paid for the program? |
| Purchases | Completed fills or swaps, asset debits, token receipts and costs | How many tokens were actually acquired? |
| Disposition | Closing holdings, distributions, locks and burn records | Where did the acquired inventory go? |
| Supply | Opening supply, issuance, burns and closing supply | Did net supply rise or fall? |
| Adjustments | Treasury exclusions, dead-address treatment and bridge offsets | Why does a provider's supply differ? |
| Valuation | Original quantities, price source and valuation timestamp | Is a dollar figure spending or a marked value? |
Finish the packet with a reproducible statement: “The treasury spent $45,000 acquiring 20,000 tokens, retained 5,000 of those acquisitions, and assigned 15,000 to a 55,000-token burn. After 25,000 of issuance, total supply fell by 30,000.” Keep the $1,000 of execution costs and $16,000 reserve draw beside it. Replace each hypothetical number with checked records before applying the statement to a real program.